Skip to content

Support to build CUDA backend with CUDA toolkit 11.0 #2939

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jun 26, 2020

Conversation

9prady9
Copy link
Member

@9prady9 9prady9 commented Jun 23, 2020

Description

  • Changes to support build with CUDA 11. We can still build with CUDA 10.* and 9.*.
  • Use descriptor based cusparse API for sparse blas fns

Fixes: #2936

Changes to Users

No changes are needed for existing developers, they should be still be able to build with their existing toolkit versions 10.* and 9.*

Checklist

  • Rebased on latest master
  • Code compiles
  • Tests pass (Tested with CUDA 10.2 and 9.2. I haven't run tests with CUDA 11 as of now).
  • Functions added to unified API
  • Functions documented

@9prady9 9prady9 requested a review from umar456 June 24, 2020 05:51
@9prady9 9prady9 dismissed umar456’s stale review June 24, 2020 06:11

Addressed feedback

@9prady9 9prady9 force-pushed the cuda_11 branch 2 times, most recently from 44f81ae to 82e4d36 Compare June 24, 2020 15:16
cusparseSpMv/cusparseSpMM functions use sparse and dense matrix/vector
descriptor objects as arguments. This API is introduced in CUDA 10.1 and
old API has been deprecated. It is also removed in CUDA 11.
@9prady9 9prady9 added this to the 3.7.2 milestone Jun 26, 2020
umar456
umar456 previously approved these changes Jun 26, 2020
Also, updates CUB version from 1.8.0 to 1.9.10
@9prady9 9prady9 merged commit 9bbc242 into arrayfire:master Jun 26, 2020
@9prady9 9prady9 deleted the cuda_11 branch June 26, 2020 15:55
@umar456 umar456 mentioned this pull request Jun 27, 2020
2 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[Build] doesn't build on CUDA 11
2 participants